Pricingby GlyphLLM price changes
Nov 2026
Embed 5 Fast
Cohere cut its API price: input $0.12 to $0.08 per 1M tokens.
GLM 5.3 FlashX
Z.ai cut its API price: cached input $0.09 to $0.075 per 1M tokens.
Oct 2026
GPT-4.1 Nano
OpenAI raised its API price: cached input $0.025 to $0.03 per 1M tokens.
Claude Sonnet 5.5
Anthropic cut its API price: cached input $0.2 to $0.1 per 1M tokens.
OpenAI raised its API price: input $2 to $4 and output $10 to $20 per 1M tokens.
Mistral Large 4
Mistral AI raised its API price: input $0.68 to $1.36 and output $2.09 to $4.18 per 1M tokens.
Grok 4.7
xAI raised its API price: input $1.6 to $2 and output $4.8 to $6 per 1M tokens.
Aug 2026
Gemini 3.7 Flash
Google raised its API price: input $0.375 to $0.75 and output $1.88 to $3.75 per 1M tokens.
OpenAI cut its API price: input $2.5 to $2 and output $15 to $10 per 1M tokens.
OpenAI cut its API price: input $5 to $2.5 and output $30 to $15 per 1M tokens.
Google cut its API price: input $1.5 to $0.75 and output $7.5 to $3.75 per 1M tokens.
Jul 2026
Qwen3.6 Max Preview
Qwen (Alibaba) cut its API price: input $1.04 to $1.03 and output $6.24 to $6.16 per 1M tokens.
OpenAI cut its API price: input $1.25 to $1 and output $7.5 to $6 per 1M tokens.
OpenAI raised its API price: cached input $0.025 to $0.03 per 1M tokens.
xAI cut its API price: cached input $0.5 to $0.3 per 1M tokens.
OpenAI cut its API price: cached input $0.01 to $0.005 per 1M tokens.
Qwen (Alibaba) raised its API price: input $1.25 to $1.48 and output $3.75 to $4.43 per 1M tokens.
Jun 2026
xAI cut its API price: input $2 to $1.25 and output $6 to $2.5 per 1M tokens.
Qwen (Alibaba) cut its API price: input $0.4 to $0.32 and output $1.6 to $1.28 per 1M tokens.
May 2026
OpenAI cut its API price: cached input $0.03 to $0.025 per 1M tokens.
Qwen (Alibaba) cut its API price: input $2.5 to $1.25 and output $7.5 to $3.75 per 1M tokens.
Qwen (Alibaba) cut its API price: input $0.25 to $0.188 and output $1.5 to $1.13 per 1M tokens.
xAI cut its API price: input $2 to $1.25 and output $6 to $2.5 per 1M tokens.
Apr 2026
Qwen (Alibaba) cut its API price: input $1.3 to $1.04 and output $7.8 to $6.24 per 1M tokens.
OpenAI raised its API price: cached input $0.005 to $0.01 per 1M tokens.
Mar 2026
Z.ai raised its API price: input $0.96 to $1.2 and output $3.2 to $4 per 1M tokens.
Qwen (Alibaba) cut its API price: input $0.4 to $0.26 and output $1.2 to $0.78 per 1M tokens.
Qwen (Alibaba) cut its API price: input $0.4 to $0.26 and output $1.2 to $0.78 per 1M tokens.
Google raised its API price: input $0.25 to $0.5 and output $1.5 to $3 per 1M tokens.
Feb 2026
Qwen (Alibaba) cut its API price: cached input $0.16 to $0.08 per 1M tokens.
Jan 2026
Google cut its API price: cached input $0.31 to $0.125 per 1M tokens.
Oct 2025
Google cut its API price: cached input $0.31 to $0.125 per 1M tokens.
Jun 2025
OpenAI cut its API price: input $10 to $2 and output $40 to $8 per 1M tokens.
Apr 2025
Anthropic raised its API price: cached input Free to $0.03 per 1M tokens.
OpenAI raised its API price: cached input Free to $1.25 per 1M tokens.
Cohere raised its API price: input $2.38 to $2.5 and output $9.5 to $10 per 1M tokens.
Oct 2024
OpenAI cut its API price: input $5 to $2.5 and output $15 to $10 per 1M tokens.
Cohere cut its API price: input $2.5 to $2.38 and output $10 to $9.5 per 1M tokens.
Sep 2024
Mistral AI cut its API price: input $3 to $2 and output $9 to $6 per 1M tokens.
Mistral AI cut its API price: input $8 to $3 and output $24 to $9 per 1M tokens.
Apr 2024
OpenAI cut its API price: input $1 to $0.5 and output $2 to $1.5 per 1M tokens.
Other recorded changes
- Kimi K3
list price: batch input price $1.084 to $0.917
listSource → - Qwen3.8 27B
primeintellect/fp8: context window 131,072 tokens to 262,144 tokens
primeintellect/fp8Source → - GLM 5.3
Route no longer offered: tenstorrent/fp8
tenstorrent/fp8Source → - GLM 5.3
Route no longer offered: nebius/fp4
nebius/fp4Source → - Kimi K3
list price: batch input price $2.28 to $1.084
listSource → - Kimi K3
list price: batch output price $11.4 to $6
listSource → - GLM 5.3 Flash
Route no longer offered: fireworks-ai/us
fireworks-ai/usSource → - DeepSeek V4.1 Flash
Route offered again: relace/fast
relace/fastSource → - DeepSeek V4.1 Flash 1017
Route offered again: relace/fast
relace/fastSource → - DeepSeek V4.1 Flash 1017
list price: batch input price none to $0.011
listSource → - DeepSeek V4.1 Flash 1017
list price: batch output price none to $0.2
listSource → - Gemma 4 31B
New route: primeintellect/fp4 at $0.4 in, $0.67 out
primeintellect/fp4Source → - DeepSeek V4 Flash 0423
Route no longer offered: qwen/fp8-us-east-1
qwen/fp8-us-east-1Source → - Kimi K3
Route no longer offered: fireworks-ai/spot
fireworks-ai/spotSource → cohere
5 new routes, 5 models
- DeepSeek V4 Flash 0731
Route no longer offered: cohere
cohereSource → - DeepSeek V4 Pro 0813
Route no longer offered: ionstream
ionstreamSource → - Qwen3.8 27B
list price: batch input price none to $0.16
listSource → - Qwen3.8 27B
list price: batch output price none to $2
listSource → - GLM 5.3 Flash
Route no longer offered: cohere
cohereSource → - DeepSeek V4.1 Flash
Route no longer offered: cohere
cohereSource → - Nemotron 3.5 Super VL
Route offered again: primeintellect/nvfp4
primeintellect/nvfp4Source → - DeepSeek V4.1 Flash 1017
Route no longer offered: cohere
cohereSource → - Kimi K3.1 Pro
New route: digitalocean at $2.0 in, $10.0 out
digitaloceanSource → - Qwen3 Embedding 8B
New route: nebius/bf16 at $0.01 in, $0.0 out
nebius/bf16Source → - Qwen3 Embedding 8B
Route no longer offered: nebius
nebiusSource → - gpt-oss-120b
Route no longer offered: mara
maraSource → - Kimi K2.6
Route no longer offered: gmicloud/fp8
gmicloud/fp8Source → - DeepSeek V4 Flash 0731
Route offered again: morph/bf16
morph/bf16Source → - DeepSeek V4 Flash 0731
nebius/fp8: context window 1,024,000 tokens to 1,048,576 tokens
nebius/fp8Source → - DeepSeek V4 Flash 0731
Route offered again: nebius/fp8
nebius/fp8Source → - Qwen3.8 27B
Route no longer offered: tenstorrent/fp4
tenstorrent/fp4Source → - GLM 5.3
Route no longer offered: modelrun/fp4
modelrun/fp4Source → - DeepSeek V4.1 Flash 1017
New route: primalabs at $0.9 in, $3.6 out
primalabsSource → - DeepSeek V4.1 Flash 1017
New route: baseten/fast at $0.6 in, $2.4 out
baseten/fastSource → - DeepSeek V4.1 Flash 1017
New route: decart/fp8 at $0.15 in, $0.6 out
decart/fp8Source → - DeepSeek V4.1 Flash 1017
Route no longer offered: decart/fp4
decart/fp4Source → - Kimi K3.1 Pro
New route: phala/us at $2.0 in, $10.0 out
phala/usSource → - Kimi K3.1 Pro
New route: wafer/us at $2.0 in, $10.0 out
wafer/usSource → - Kimi K3.1 Pro
New route: fireworks-ai at $2.0 in, $10.0 out
fireworks-aiSource → - Kimi K3.1 Pro
New route: primalabs/mxfp4 at $1.8 in, $9.0 out
primalabs/mxfp4Source → - Kimi K3.1 Pro
New route: sail-research/fp4 at $0.346 in, $16.581 out
sail-research/fp4Source → - Kimi K3.1 Pro
New route: sail-research/fp4-us at $0.346 in, $16.581 out
sail-research/fp4-usSource → - Mixtral 8x22B Instruct
Route no longer offered: list price
listSource → parasail
5 new routes, 5 models
- gpt-oss-120b
New route: akashml/fp4 at $0.03 in, $0.187 out
akashml/fp4Source → - gpt-oss-120b
Route no longer offered: akashml/bf16
akashml/bf16Source → - Voxtral Small 24B 2507
Route no longer offered: list price
listSource → - Ministral 3 3B 2512
Route no longer offered: list price
listSource → - Ministral 3 8B 2512
Route no longer offered: list price
listSource → - Ministral 3 14B 2512
Route no longer offered: list price
listSource → - Mistral Small 4
Route no longer offered: list price
listSource → - DeepSeek V4 Flash 0423
Route no longer offered: crusoe/fp8
crusoe/fp8Source → - DeepSeek V4 Pro 0423
Route no longer offered: crusoe/fp8
crusoe/fp8Source → - Kimi K3
Route offered again: parasail/fast
parasail/fastSource → - Kimi K3
New route: io-net/fp4 at $2.3 in, $14.0 out
io-net/fp4Source → - Kimi K3
Route no longer offered: akashml/fp4
akashml/fp4Source → - Qwen3.8 2.4T A95B
Route no longer offered: together-ai
together-aiSource → - Qwen3.8 27B
New route: nebius/fp4 at $0.45 in, $3.0 out
nebius/fp4Source → - GLM 5.3
New route: coreweave/nvfp4 at $0.75 in, $3.0 out
coreweave/nvfp4Source → - GLM 5.3
Route no longer offered: coreweave/fp8
coreweave/fp8Source → - GLM 5.3
Route no longer offered: mistral-ai/nvfp4
mistral-ai/nvfp4Source → - GLM 5.3 Flash
Route offered again: parasail/fast
parasail/fastSource → - GLM 5.3 Flash
New route: makora/fp4 at $0.14 in, $0.24 out
makora/fp4Source → - GLM 5.3 Flash
Route no longer offered: makora
makoraSource → - DeepSeek V4.1 Flash
Route no longer offered: sail-research/fp4-us
sail-research/fp4-usSource → - DeepSeek V4.1 Flash
Route no longer offered: sail-research/fp4
sail-research/fp4Source → - Mistral Large 4
Route no longer offered: list price
listSource → - Nemotron 3.5 Super VL
Route offered again: nebius/bf16
nebius/bf16Source → - DeepSeek V4.1 Flash 1017
New route: coreweave/fp8 at $0.3 in, $1.2 out
coreweave/fp8Source → inferencenet
4 new routes, 1 model
- Kimi K3.1 Pro
New route: qwen at $2.3 in, $11.5 out
qwenSource → - Kimi K3.1 Pro
New route: together-ai at $2.0 in, $10.0 out
together-aiSource → - Kimi K3.1 Pro
New route: nebius/fp4 at $2.0 in, $10.0 out
nebius/fp4Source → - Kimi K3.1 Pro
New route: phala at $2.0 in, $10.0 out
phalaSource → - Kimi K3.1 Pro
New route: modal/mxfp4 at $2.0 in, $10.0 out
modal/mxfp4Source → - Kimi K3.1 Pro
New route: akashml/fp4 at $2.0 in, $10.0 out
akashml/fp4Source → - Kimi K3.1 Pro
New route: morph/fp8 at $1.908 in, $12.0 out
morph/fp8Source → - Kimi K3.1 Pro
New route: deepinfra/bf16 at $2.0 in, $10.0 out
deepinfra/bf16Source → - Kimi K3.1 Pro
New route: wafer/mxfp4 at $1.8 in, $9.0 out
wafer/mxfp4Source → - Kimi K3.1 Pro
New route: relace/fp4 at $0.4 in, $7.8833 out
relace/fp4Source → - Kimi K3.1 Pro
Route no longer offered: list price
listSource → novita-ai
4 routes no longer offered, 4 models
- Llama 3.1 70B Instruct
Route no longer offered: deepinfra/turbo
deepinfra/turboSource → - DeepSeek V3.1
Route no longer offered: mara
maraSource → - DeepSeek V3.2
Route no longer offered: sambanova
sambanovaSource →
Sources: the daily 05:00 UTC read of OpenRouter's catalog, and Internet Archive snapshots before tracking began. A change is dated by the first read that showed the new price; the move happened between that read and the one before. A move of 2x or more, or a second move the same way within 7 days, waits unverified until it is checked against the lab's own page; a price that depends on the hour (DeepSeek's peak hours) is recorded at the rate in force most of the week.